pathlib mistakenly assumes os.getcwd() is a resolved path in Windows

[eryksun]

BPO	43455
Nosy	@pfmoore, @tjguk, @zware, @eryksun, @zooba, @uranusjr
PRs	bpo-38671: Make sure to return an absolute path in pathlib._WindowsFlavour.resolve() #17716

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = None
created_at = <Date 2021-03-09.23:05:56.225>
labels = ['type-bug', '3.8', '3.9', '3.10', 'library', 'OS-windows']
title = 'pathlib mistakenly assumes os.getcwd() is a resolved path in Windows'
updated_at = <Date 2022-03-03.10:05:28.651>
user = 'https://github.com/eryksun'

bugs.python.org fields:

activity = <Date 2022-03-03.10:05:28.651>
actor = 'uranusjr'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = ['Library (Lib)', 'Windows']
creation = <Date 2021-03-09.23:05:56.225>
creator = 'eryksun'
dependencies = []
files = []
hgrepos = []
issue_num = 43455
keywords = ['patch']
message_count = 2.0
messages = ['388393', '414415']
nosy_count = 6.0
nosy_names = ['paul.moore', 'tim.golden', 'zach.ware', 'eryksun', 'steve.dower', 'uranusjr']
pr_nums = ['17716']
priority = 'normal'
resolution = None
stage = 'patch review'
status = 'open'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue43455'
versions = ['Python 3.8', 'Python 3.9', 'Python 3.10']

[eryksun]

pathlib._WindowsFlavour.resolve() mistakenly assume that os.getcwd() returns a resolved path in Windows:

    s = str(path)
    if not s:
        return os.getcwd()

I don't think this is a practical problem since str(path) should never be an empty string. But if there is a concern that the result is an empty string, the code should use s = str(path) or '.', and resolve "." like any other relative path.

In POSIX the result of getcwd() "shall contain no components that are dot or dot-dot, or are symbolic links". In Windows, os.getcwd() calls WinAPI GetCurrentDirectoryW(), which returns a fully-qualified path that may contain symbolic components that would be resolved in a final path. This includes filesystem symlinks and bind mounts (junctions), as well as mapped and substitute drives (i.e. drives that resolve to a filesystem directory instead of a volume device).

[uranusjr]

I believe this has been fixed in #25264

[barneygale]

Hey @eryksun, could you check whether this is valid after #25264 was merged? Pathlib now calls ntpath.realpath(), which (from scanning the code) doesn't appear to suffer from this bug.

[eryksun]

I never had a practical case for not s, since it's at least ".". Anyway, this is a non-issue now. It was only kept open for 3.8 and 3.9, and the last bugfix for 3.9 was released on 2022-05-17.